Buoy identification and positioning method based on NPU board card and shipborne device thereof

Abstract

The invention provides a buoy identification and positioning method based on an NPU board card and a shipborne device thereof, and the method comprises the following steps: binding a buoy on a ship through a cable, pulling the buoy to the sea, aligning N groups of cameras with the buoy for photographing, and obtaining a multi-dimensional diversified buoy image sample; turning on an NPU board cardpower supply, and importing model parameters of the convolutional neural network; starting a camera to output a video stream, and transmitting the video stream to the NPU board card; after the detection control program acquires the video stream, analyzing a picture from the video stream, calling a target detection model based on a convolutional neural network, and identifying the buoy type and thepixel position in the image; and packaging the target identification result list into a data packet, and sending the data packet to an identification result client. According to the buoy identifyingand positioning method based on the NPU board card and the shipborne device thereof provided by the invention, the technical problem that the buoy cannot be accurately and automatically identified andpositioned in real time on a ship is solved, so that the ship can quickly detect and avoid the buoy in time.

Description

technical field [0001] The invention relates to the technical fields of marine monitoring and artificial intelligence, in particular to an NPU board-based buoy identification and positioning method and a shipboard device thereof. Background technique [0002] Buoy recognition and positioning, also known as buoy target detection, is to accurately find the location of the object in a given picture and mark the category of the object. Therefore, the problem to be solved by buoy target detection is the whole process of where and what the object is. However, this problem is not easy to solve because the size of the object varies widely, the angle and pose of the object are not fixed, and it can appear anywhere in the picture. In addition, the object may still have multiple categories. At present, the buoy target detection algorithms emerging in academia and industry are divided into three categories: [0003] 1) Traditional target detection algorithm: cascade Cascade + HOG / DPM ...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to provide a buoy identification and positioning method based on an NPU board and its ship-borne device, which solves the technical problem that the radar and other sensor devices cannot be used to accurately and automatically identify and locate the buoy in the past, so that the ship can Quickly detect and avoid buoys in time

Images